Statistical discrepancy

Sometimes there is a question about some discrepancy in statistics in the advertising network, and, for example, an affiliate program or your website. If the discrepancy in statistics is not significant (10-20% of all transitions), then this may occur due to:

The counter of pages to which traffic is sent is usually located in the footer of the page, and is loaded last - often the visitor leaves it without waiting for the entire page to load. Therefore, the statistics of the third-party service does not count the transition, although it was. There is also a situation when not all services count proxy traffic as unique. There are also interruptions in the work of the site, or a third-party counter. That's about transitions.

A discrepancy in the geo location of the user may arise due to the difference in the bases by which the visitor is identified. As you know, there is no 100% way to accurately determine the user's location. We use the most accurate baseline we've tested.

The main characteristic of whois is the content of information about the location of the subnet owner, not about the geographic location of the network itself. And even these data are often inaccurate, sometimes deliberately false. We will describe just a few cases with examples of how the lack of providing just such information can affect the correct determination of a person's location.

First, this applies to global telecommunications giants, hosting companies and data centers. A small example: the network 67.15.107.64 - 67.15.107.127 in whois is represented as Uruguayan. It is possible that the firm's office is located in Uruguay, but the servers for these IP-addresses are located on the territory of ThePlanet data center in Houston (Houston, USA).

Large companies with offices around the world do not share information about their real location through whois data.

  • Example of this: Google Inc. The company's data centers are located all over the world, but whois always indicates the head office in Mountain View (California, USA). In reality, users from different countries are directed to the nearest data center. For example, for a resident of Germany, the main page will open from the address 74.125.39.104, a German data center.
  • The same for small firms: the whois indicates the location of the main office, no one will remember the rest of the offices located around the world. For example, the network 12.174.253.208-12.174.253.255 is designated in whois as US, but in reality it is a branch of the firm in Puerto Rico (PR).

There are also common human errors:

  • Incorrectly specified range, example: 202.154.32.0 - 202.154.64.63 in whois is designated as Australia (AU), but it should be 202.154.64.0 - 202.154.64.63. You can notice this error, for example, by looking at the address 202.154.37.1
  • The usual mistakes in the name of the country: the network 92.42.250.0 - 92.42.250.3 should be like Serbia (RS) instead of Suriname (SR).
  • Use of different abbreviations that match different country. For example, one Ukrainian hosting for some clients and their subnets indicates "NA". Perhaps this is the abbreviation for "not available". But the servers are not in Namibia :))) 78.109.16.24 - 78.109.16.31
  • "Dead" web. For example, in the case of 198.133.81.0-198.133.85.255, owned by Fisons, which was acquired in 1997. In fact, now this range no longer exists, in whois it was simply not deleted.

These are the most common cases and cover a fairly large percentage of IP addresses around the world. Realizing this, we look at things realistically, and unlike many other ad networks, we guarantee at least 60% of the ordered traffic by geo-tag.

BUT! If the discrepancy in traffic is too large (both by clicks and by geo-basis), this may indicate any malfunctions in the system, or cheating, and serve as a reason for a detailed checks. Contact support to understand the situation in detail and find out the true reasons.

24 мая 2021 г.